Background. There has been a 30% increase in chronic kidney disease (CKD) over the past decade. Anemia almost invariably develops in patients with progressive chronic renal insufficiency, and is associated with a number of adverse outcomes. Erythropoiesis-stimulating agents (ESA therapy) - approved in 1989 for use among the renal disease population - is used to correct anemia. When to start ESA therapy - at what hematocrit level - and at what threshold level to maintain hematocrit throughout treatment is unknown. For example, should ESA therapy be initiated when patients are severely anemic (i.e., hematocrit below 30%) or only moderately anemic (i.e., hematocrit between 30 - 33%)? Even more fundamental, providers appear to be uncertain about the clinical benefits to be derived from epoetin therapy for their anemic CKD patients. Objective. We will examine three possible treatment strategies to address the following research question: Do different anemia management treatment strategies among the CKD population result in different patient outcomes with regard to progression of CKD, cardiovascular disease, and mortality? The strategies that will be examined in this project include: 1) ESAs are not prescribed for patients with anemia; 2) ESAs are prescribed for severely anemic patients (hct < 30%); and 3) ESAs are prescribed for moderately anemic patients (hct between 30 - 33%). Methods. We propose to use an observational study using innovative causal modeling techniques that 'mimics' RCTs by appropriately handling time-dependent confounding between therapy and intermediate clinical outcomes. Specifically, patients who do not follow one of the three strategies listed above throughout the study will generate time- dependent selection bias when conventional statistical techniques are used. Instead, Marginal Structural Models (MSMs) can handle such biases with the caveat that treatment-to-treatment data (including ESA dosing information and hematocrit values) must be used in the modeling process. We propose to use Kaiser Permanente Mid-Atlantic data containing treatment to treatment information as well as detailed laboratory data (GFR, creatinine, iron) and medication use. Significance. Notably, despite the lack of consensus for the use of ESAs, and the large number of CKD patients who never receive therapy, the volume of ESAs administered to predialysis patients has increased 5-fold in the last decade. Our application - to examine different strategies for initiating and maintaining ESA therapy - has important implications for both anemia management and its costs among the burgeoning CKD population.